REPLY TO “MAXIMAL VIOLATION OF BELL INEQUALITIES BY POSITION MEASUREMENTS” DANIEL V. TAUSK Abstract. In a recent article [9], Kiukas and Werner claim to have shown that Bohmian Mechanics does not make the same empirical pre- dictions as ordinary Quantum Mechanics. More precisely, they have shown that ordinary Quantum Mechanics predicts maximal violations of the CHSH–Bell inequality for a certain experiment in which only po- sition measurements are performed on two noninteracting entangled free non-relativistic particles. Kiukas and Werner claim that Bohmian Me- chanics doesn’t predict a violation of the CHSH–Bell inequality for that experiment. We explain that such claim is wrong and that the argu- ment supporting their claim neglects the fact that Bohmian Mechanics does not satisfy all the assumptions needed to prove the CHSH–Bell in- equality. We also clear up another few misconceptions about Bohmian Mechanics appearing in [9]. The CHSH inequality1 concerns an experiment in which, after a prepara- tion procedure, two experimenters (normally referred to as Alice and Bob) can each choose between two options for running an experiment having two possible outcomes, to which one assign the numbers 1 and +1. The ex- periments are assumed to be performed at spacelike separati− on. Denoting Alice’s outcome by A 1, 1 and Bob’s outcome by B 1, 1 , then the CHSH inequality reads:∈ {− } ∈ {− } E11(AB)+ E12(AB)+ E21(AB) E22(AB) 2, | − |≤ where Eab denotes expected value and a, b 1, 2 refer respectively to Alice and Bob’s possible experimental choices.∈ { The} CHSH inequality can either be proved from Bell’s locality condition2 or from the assumption of the arXiv:1408.2900v1 [quant-ph] 13 Aug 2014 existence of so called non-contextual hidden variables. In the present con- text, the latter assumption means that A and B can be written as functions A = A(a, λ), B = B(b, λ) where, as above, a, b 1, 2 denote the possible experimental choices, and λ Λ denotes a random∈ { parameter} belonging to a probability space Λ. It is∈ crucial for the proof of the CHSH inequality that A not be allowed to depend on b and B not be allowed to depend on Date: December 7th, 2010. 1After Clauser, Horne, Shimony and Holt [6]. The inequality is also sometimes referred to as “Bell’s inequality”, even though it is not the inequality appearing in Bell’s celebrated paper [1]. Nevertheless, Bell himself used the CHSH inequality to explain his nonlocality argument in several papers (see [5]). 2See [2, 3] and also [4, 10] for further discussion on Bell’s locality condition. 1 REPLY TO “MAXIMAL VIOLATION OF BELL INEQUALITIES” 2 a. It is precisely the fact that Bohmian Mechanics does not satisfy such as- sumption that Kiukas and Werner have overlooked. If Alice’s (resp., Bob’s) experimental choices are associated to two 1, 1 -valued quantum observ- {− } ables represented by self-adjoint operators on a Hilbert space A (resp., H B) having spectrum contained in 1, 1 and the preparation procedure H {− } is associated to a quantum state on the tensor product A B then Quantum Theory predicts that the CHSH correlation (theH lefthand⊗ H side of the CHSH inequality) is bounded by 2√2. In their elegant paper [9], Kiukas and Werner present efficient techniques for determining conditions under which there exist states that maximally violate the CHSH inequality (i.e., for which the CHSH correlation equals 2√2) as well as for explicitly deter- mining such states. Such techniques are then applied to obtain quantum states that maximally violate the CHSH inequality for two noninteracting entangled free non-relativistic particles, where the quantum observables cor- responding to the experimenters’ choices are 1, 1 -valued functions of the Heisenberg picture position operator at two distinct{− } times. In other words, one obtains a maximal violation of the CHSH inequality by position mea- surements alone3. 3 In Bohmian Mechanics, particles have trajectories t Qk(t) R which satisfy the first order ordinary differential equation: 7→ ∈ ∗ dQk ~ ψ kψ (1) (t)= Im ∇∗ t,Q1(t),...,Qn(t), k = 1,...,n dt mk ψ ψ where ψ denotes the wave function satisfying the standard Schr¨odinger equa- tion and mk the mass of the k-th particle. Let QA and QB denote, respec- tively, the Bohmian trajectories of Alice’s and of Bob’s particle. Let us assume that, after the preparation of the initial wave function, both par- ticles are free (i.e., there is no potential term in Schr¨odinger’s equation and no interaction with a measurement apparatus). If one defines Alice’s (resp., Bob’s) outcome A (resp., B) as being the appropriate 1, 1 -valued {− } function of the Bohmian position QA(ta) (resp., QB(tb)) at the appropri- ate instant ta (resp., tb) that depends on Alice’s (resp., Bob’s) experimental choice a 1, 2 (resp., b 1, 2 ) then the CHSH inequality is not violated: namely,∈ if {one takes} the random∈ { } parameter λ to be the Bohmian positions QA, QB at an instant right after the preparation of the initial wave function then A is indeed a function of a and λ (similarly, B is a function of b and λ), so that all the assumptions needed to prove the CHSH inequality are satisfied. However, such observation is completely irrelevant for the issue of comparing the empirical predictions of Bohmian Mechanics with those of ordinary Quantum Mechanics: namely, ordinary Quantum Mechanics has nothing to say about what happens to particles when they are left unob- served. Ordinary Quantum Mechanics makes predictions for what Alice and 3The hardest part of the work is the proof of the existence of maximally violating states. It is easy to prove the existence of states violating CHSH from the assumption that Alice’s and Bob’s observables do not commute (see formula (3) of [9]). REPLY TO “MAXIMAL VIOLATION OF BELL INEQUALITIES” 3 Bob will see if they use a particle detector and that is the experiment that one must analyze under Bohmian Mechanics in order to make a comparison. It turns out that, once the interaction of the particles with the detectors are taken into account, Bohmian Mechanics does not satisfy the assumptions needed to prove the CHSH inequality as the interaction of a particle with a detector influences not only the Bohmian trajectory of that particle but also the Bohmian trajectory of the distant particle (assuming that the particles are entangled), so that now Bob’s outcome B is a function of λ and of both Alice’s choice a and Bob’s choice b4. It turns out that Bohmian Mechanics makes the same experimental predictions as ordinary Quantum Mechanics for that experiment5 and that Bohmian Mechanics predicts a (maximal) violation of CHSH when ordinary Quantum Mechanics does so. While the formulation of Bohmian Mechanics does not mention any “wave function collapse” and the wave function of a Bohmian universe evolves ex- clusively through the Schr¨odinger equation, it turns out that, as a mathe- matical consequence of the dynamical equations of the theory, the wave func- tion of a subsystem of a Bohmian universe6 does not always evolve through the Schr¨odinger equation, but it sometimes goes through the standard text- book “collapse under measurement” (see [11, 8] for a succinct exposition and [7] for a more detailed one). According to Kiukas and Werner ([9, Section II]): “The simplest position [the proponents of Bohmian Mechan- ics can adopt in order to make the agreement with ordinary Quantum Mechanics possible] is to include the collapse of the wave function into the theory. Then the first measurement instantaneously collapses the wave function. So if agreement with quantum mechanics is to be kept, the probability dis- tribution changes suddenly. There is no way to fit this with continuous trajectories: When the guiding field collapses, the particles must jump.” 4Since, according to Bohmian Mechanics, Bob’s outcome B depends on Alice’s choice a, one might get the impression that Bohmian Mechanics predicts that Alice can send superluminal messages to Bob. However, just like in ordinary Quantum Mechanics, such superluminal messaging is not possible. In order to “read” Alice’s message (the a), Bob would have to know more about the value of λ — the initial Bohmian configuration — than it is possible for him to know without disturbing the preparation of the initial wave function (see [8, Sections 4.2, 4.3] for a succinct exposition and [7] for a more detailed one). 5That this is indeed the case follows, for instance, from the following observation: if t is an instant after which both experiments have been performed then the Bohmian particle configuration of the apparatuses is |ψ(t)|2-distributed and such configuration records the outcomes of both experiments. 6The wave function of a subsystem is obtained by plugging into the wave function ψ of the Bohmian universe the actual Bohmian positions of the particles not belonging to that subsystem. Some technical complications arise in the case of particles with spin which we can safely ignore for the purpose at hand. REPLY TO “MAXIMAL VIOLATION OF BELL INEQUALITIES” 4 So, contrary to Kiukas and Werner’s claim, one does not have to add any collapse rule into Bohmian Mechanics, for, as explained above, the collapse of the wave function of a subsystem follows from the standard dynamical equations of the theory. Moreover, contrary to their claim, there is no problem in fitting such collapse with continuous trajectories, as Bohmian trajectories are smooth. One should point out that the wave function of a subsystem of a Bohmian universe does not go under any instantaneous collapse: namely, the wave function of a subsystem is a smooth function of both time and configuration point and collapse happens over a positive amount of time.